I love playing around with voice assistants and built a few for different rooms. Why is it not possible to record short audio messages from a voice assistant and send it to a different room?
In kitchen: “Okay Nabu. Send audio message to living room. [Recording starts] Please come to lunch in 5 minutes.”
Living room, recorded message is played: “Please come to lunch in 5 minutes.”
I’ve actually implemented something similar in my setup, and it works like a charm! Here’s how I do it:
I have a predefined list of 24 voice messages, such as “Coffee’s ready,” “We’re leaving in 5 minutes,” or “Dinner’s served.” These messages can be triggered in a couple of ways:
Voice commands through Assist and Google AI,
Dynamic dropdown menus in Home Assistant.
The dynamic dropdown is the real game-changer. For example, in the morning, the menu might include “Breakfast is ready,” but not “Dinner is ready.” In the evening, it’s the opposite. This keeps the list concise and practical, avoiding endless scrolling to find the right message.
Next to this dropdown, there’s another menu for selecting which speakers the message should play on. If none of the predefined messages fit the situation, there’s always the “Custom Message” option at the end of the list. Selecting it reveals a text field where I can type a custom message, hit “Send,” and voilà—it plays on the selected speakers.
To make it more engaging, automations and scripts use an LLM to generate the messages in different styles, so they don’t feel repetitive.
Currently, I only have two HA Assist speakers (one on each floor), so I’m really looking forward to Nabu Casa’s upcoming speakers! In the meantime, the GUI selection ensures everything is still manageable and flexible.
This setup has been a fantastic way to integrate voice messages across the home in a way that feels seamless and user-friendly.
This can probably already be done with custom sentences using something like “announce {message}” and then an action to announce the {message} on all speakers. Although it will use TTS, not your voice.
I think someone was working on a walkie-talkie type function in ESP home…
I would love to have the option to send a raw voice recording to a specified speaker in the house. Or a trigger word which established a voice socket between two or more endpoints. A virtual intercom built on top of the voice assistant hardware platform.
This way I could quickly send a message 3 floors up, hey go get the eggs, the sun is going down and the chickens are going to bed, hurry up.